N-gram based approach to automatic Tamil lyric generation by identifying emotion

Abstract

This paper discusses a tri-gram approach to automatic Tamil lyric generation. The approach is based on identifying the emotion from a given scenario and uses this emotion as a seed word to interpret the context of the scenario. A lyric model based on tri-gram is constructed which is referred using the identified seed word to generate lyrics. The lyric model, tri-gram of words, Tamil sentence rules and suffixes are used by a Morphological generator to generate lyrics. Using this approach we achieved an average accuracy of 74.17% with respect to exact emotion being conveyed in the generated lyrics. © 2013 Springer.